Thrombosis—the formation, development, or presence of a blood clot (thrombus) in a blood vessel—is the most common severe medical disorder. The most frequent example of arterial thrombosis is coronary thrombosis, which leads to occlusion of the coronary arteries and often to myocardial infarction (heart attack). More than 1.3 million patients are admitted to the hospital for myocardial infarction each year in North America. The standard therapy is administration of a thrombolytic protein by infusion. Thrombolytic treatment of acute myocardial infarction is estimated to save 30 lives per 1000 patients treated; nevertheless the 30-day mortality for this disorder remains substantial (Mehta et al., Lancet 356:449-454 (2000), incorporated herein by reference). It would be convenient to administer antithrombotic and thrombolytic agents by bolus injection, since they might be used before admission to hospital with additional benefit (Rawles, J. Am. Coll. Cardiol. 30:1181-1186 (1997), incorporated herein by reference). However, bolus injection (as opposed to a more gradual intravenous infusion) significantly increases the risk of cerebral hemorrhage (Mehta et al., 2000). The development of an agent able to prevent thrombosis and/or increase thrombolysis, without augmenting the risk of bleeding, would be desirable.
Unstable angina, caused by inadequate oxygen delivery to the heart due to coronary occlusion, is the most common cause of admission to hospital, with 1.5 million cases a year in the United States alone. When patients with occlusion of coronary arteries are treated with angioplasty and stenting, the use of an antibody against platelet gp IIb/IIIa decreases the likelihood of restenosis. However, the same antibody has shown no benefit in unstable angina without angioplasty, and a better method for preventing coronary occlusion in these patients is needed.
Another important example of arterial thrombosis is cerebral thrombosis. Intravenous recombinant tissue plasminogen activator (rtPA) is the only treatment for acute ischemic stroke that is approved by the Food and Drug Administration. The earlier it is administered the better (Ernst et al., Stroke 31:2552-2557 (2000), incorporated herein by reference). However, intravenous rtPA administration is associated with increased risk of intracerebral hemorrhage. Full-blown strokes are often preceded by transient ischemic attacks (TIA), and it is estimated that about 300,000 persons suffer TIA every year in the United States. It would be desirable to have a safe and effective agent that could be administered as a bolus and would for several days prevent recurrence of cerebral thrombosis without increasing the risk of cerebral hemorrhage. Thrombosis also contributes to peripheral arterial occlusion in diabetics and other patients, and an efficacious and safe antithrombotic agent for use in such patients is needed.
Venous thrombosis is a frequent complication of surgical procedures such as hip and knee arthroplasties. It would be desirable to prevent thrombosis without increasing hemorrhage into the field of operation. Similar considerations apply to venous thrombosis associated with pregnancy and parturition. Some persons are prone to repeated venous thrombotic events and are currently treated by antithrombotic agents such as coumarin-type drugs. The dose of such drugs must be titrated in each patient, and the margin between effective antithrombotic doses and those increasing hemorrhage is small. Having a treatment with better separation of antithrombotic activity from increased risk of bleeding is desirable. All of the recently introduced antithrombotic therapies, including ligands of platelet gp IIb/IIIa, low molecular weight heparins, and a pentasaccharide inhibitor of factor Xa, carry an increased risk of bleeding (Levine et al., Chest 119:108S-121S (2001), incorporated herein by reference). Hence there is a need to explore alternative strategies for preventing arterial and venous thrombosis without augmenting the risk of hemorrhage.
To inhibit the extension of arterial or venous thrombi without increasing hemorrhage, it is necessary to exploit potential differences between mechanisms involved in hemostasis and those involved in thrombosis in large blood vessels. Primary hemostatic mechanisms include the formation of platelet microaggregates, which plug capillaries and accumulate over damaged or activated endothelial cells in small blood vessels. Inhibitors of platelet aggregation, including agents suppressing the formation or action of thromboxane A2, ligands of gp IIa/IIIb, and drugs acting on ADP receptors such as clopidogrel (Hallopeter, Nature 409:202-207 (2001), incorporated herein by reference), interfere with this process and therefore increase the risk of bleeding (Levine et al., 2001). In contrast to microaggregate formation, occlusion by an arterial or venous thrombus requires the continued recruitment and incorporation of platelets into the thrombus. To overcome detachment by shear forces in large blood vessels, platelets must be bound tightly to one another and to the fibrin network deposited around them.
Evidence has accumulated that the formation of tight macroaggregates of platelets is facilitated by a cellular and a humoral amplification mechanism, which reinforce each other. In the cellular mechanism, the formation of relatively loose microaggregates of platelets, induced by moderate concentrations of agonists such as ADP, thromboxane A2, or collagen, is accompanied by the release from platelet α-granules of the 85-kD protein Gas6 (Angelillo-Scherrer et al., Nature Medicine 7:215-221 (2001), incorporated herein by reference). Binding of released Gas6 to receptor tyrosine kinases (Axl, Sky, Mer) expressed on the surface of platelets induces complete degranulation and the formation of tight macroaggregates of these cells. In the humoral amplification mechanism, a prothrombinase complex is formed on the surface of activated platelets and microvesicles. This generates thrombin and fibrin. Thrombin is itself a potent platelet activator and inducer of the release of Gas6 (Ishimoto and Nakano, FEBS Lett. 446:197-199 (2000), incorporated herein by reference). Fully activated platelets bind tightly to the fibrin network deposited around them. Histological observations show that both platelets and fibrin are necessary for the formation of a stable coronary thrombus in humans (Falk et al. Interrelationship between atherosclerosis and thrombosis. In Vanstraete et al. (editors), Cardiovascular Thrombosis: Thrombocardiology and Thromboneurology. Philadelphia: Lipincott-Raven Publishers (1998), pp. 45-58, incorporated herein by reference). Another platelet adhesion molecule, amphoterin, is translocated to the platelet surface during activation, and binds anionic phospholipid (Rouhainen et al., Thromb. Hemost. 84:1087-1094 (2000), incorporated herein by reference). Like Gas6, amphoterin could form a bridge during platelet aggregation.
The question arises whether it is possible to inhibit these amplification mechanisms but not the initial platelet aggregation step, thereby preventing thrombosis without increasing hemorrhage. The importance of cellular amplification has recently been established by studies of mice with targeted inactivation of Gas6 (Angelillo-Scherrer et al., 2001). The Gas6−/− mice were found to be protected against thrombosis and embolism induced by collagen and epinephrine. However, the Gas6−/− mice did not suffer from spontaneous hemorrhage and had normal bleeding after tail clipping. Furthermore, antibodies against Gas6 inhibited platelet aggregation in vitro as well as thrombosis induced in vivo by collagen and epinephrine. In principle, such antibodies, or ligands competing for Gas6 binding to receptor tyrosine kinases, might be used to inhibit thrombosis. However, in view of the potency of humoral amplification, it might be preferable to inhibit that step. Ideally such an inhibitor would also have additional suppressive activity on the Gas6-mediated cellular amplification mechanism.
A strategy for preventing both cellular and humoral amplification of platelet aggregation is provided by the annexins, a family of highly homologous antithrombotic proteins of which ten are expressed in several human tissues (Benz and Hofmann, Biol. Chem. 378:177-183 (1997), incorporated herein be reference). Annexins share the property of binding calcium and negatively charged phospholipids, both of which are required for blood coagulation. Under physiological conditions, negatively charged phospholipid is mainly supplied by phosphatidylserine (PS) in activated or damaged cell membranes. In intact cells, PS is confined to the inner leaflet of the plasma membrane bilayer and is not accessible on the surface. When platelets are activated, the amounts of PS accessible on their surface, and therefore the extent of annexin binding, are greatly increased (Sun et al., Thrombosis Res. 69:289-296 (1993), incorporated herein by reference). During activation of platelets, microvesicles are released from their surfaces, greatly increasing the surface area expressing anionic phospholipids with procoagulant activity (Merten et al., Circulation 99:2577-2582 (1999); Chow et al., J. Lab. Clin. Med. 135:66-72 (2000), both incorporated herein by reference). These may play an important role in the propagation of platelet-mediated arterial thrombi.
Proteins involved in the blood coagulation cascade (factors X, Xa, and Va) bind to membranes bearing PS on their surfaces, and to one another, forming a stable, tightly bound prothrombinase complex. Several annexins, including II, V, and VIII, bind PS with high affinity, thereby preventing the formation of a prothrombinase complex and exerting antithrombotic activity. Annexin V binds PS with very high affinity (Kd=1.7 nmol/L), greater than the affinity of factors X, Xa, and Va for negatively charged phospholipids (Thiagarajan and Tait, J. Biol. Chem. 265:17420-17423 (1990), incorporated herein by reference). Tissue factor-dependent blood coagulation on the surface of activated or damaged endothelial cells also requires surface expression of PS, and annexin V can inhibit this process (van Heerde et al., Arterioscl. Thromb. 14:824-830 (1994), incorporated herein by reference), although annexin is less effective in this activity than in inhibition of prothrombinase generation (Rao et al., Thromb. Res. 62:517-531 (1992), incorporated herein by reference).
The binding of annexin V to activated platelets and to damaged cells probably explains the selective retention of the protein in thrombi. This has been shown in experimental animal models of venous and arterial thrombosis (Stratton et al., Circulation 92:3113-3121 (1995); Thiagarajan and Benedict, Circulation 96:2339-2347 (1997), both incorporated herein by reference), and labeled annexin has been proposed for medical imaging of vascular thrombi in humans, with reduced noise and increased safety (Reno and Kasina, International Patent Application PCT/US95/07599 (WO 95/34315) (published Dec. 21, 1995), incorporated herein by reference). The binding to thrombi of a potent antithrombotic agent such as annexin V provides a strategy for preventing the extension or recurrence of thrombosis. Transient myocardial ischemia also increases annexin V binding (Dumont et al., Circulation 102:1564-1568 (2000), incorporated herein by reference). Annexin V imaging in humans has shown increased binding of the protein in transplanted hearts when endomyocardial biopsy has demonstrated vascular rejection (Acio et al., J. Nuclear Med. 41 (5 Suppl.):127P (2000), incorporated herein by reference). This binding is presumably due to PS exteriorized on the surface of damaged endothelial cells, as well as of apoptotic myocytes in hearts that are being rejected. It follows that administration of annexin after myocardial infarction should prevent the formation of pro-thrombotic complexes on both platelets and endothelial cells, thereby preventing the extension or recurrence of thrombosis. Annexin V binding is also augmented following cerebral hypoxia in humans (D'Arceuil et al., Stroke 2000: 2692-2700 (2000), incorporated herein by reference), which supports the hypothesis that administration of annexin following TIA may decrease the likelihood of developing a full-blown stroke.
Annexins have shown anticoagulant activity in several in vitro thrombin-dependent assays, as well as in experimental animal models of venous thrombosis (Romisch et al., Thrombosis Res. 61:93-104 (1991); Van Ryn-McKenna et al., Thrombosis Hemostasis 69:227-230 (1993), both incorporated herein by reference) and arterial thrombosis (Thiagarajan and Benedict, 1997). Remarkably, annexin in antithrombotic doses had no demonstrable effect on traditional ex vivo clotting tests in treated rabbits (Thiagarajan and Benedict, 1997) and did not significantly prolong bleeding times of treated rats (Van Ryn-McKenna et al., 1993). In treated rabbits annexin did not increase bleeding into a surgical incision (Thiagarajan and Benedict, 1997). Thus, uniquely among all the agents so far investigated, annexins exert antithrombotic activity without increasing hemorrhage. Annexins do not inhibit platelet aggregation triggered by agonists other than thrombin (van Heerde et al., 1994), and platelet aggregation is the primary hemostatic mechanism. In the walls of damaged blood vessels and in extravascular tissues, the tissue factor/VIIa complex also exerts hemostatic effects, and this system is less susceptible to inhibition by annexin V than is the prothrombinase complex (Rao et al., 1992). This is one argument for confining administered annexin V to the vascular compartment as far as possible; the risk of hemorrhage is likely to be reduced.
Despite such promising results for preventing thrombosis, a major problem associated with the therapeutic use of annexins is their short half-life in the circulation, estimated in experimental animals to be 5 to 15 minutes (Römisch et al., 1991; Stratton et al., 1995; Thiagarajan and Benedict, 1997); annexin V also has a short half-life in the circulation of humans (Strauss et al., J. Nuclear Med. 41 (5 Suppl.):149P (2000), incorporated herein by reference). Most of the annexin is lost into the urine, as expected of a 36 kDa protein (Thiagarajan and Benedict, 1997). There is a need, therefore, for a method of preventing annexin loss from the vascular compartment into the extravascular compartment and urine, thereby prolonging antithrombotic activity following a single injection.